The development and improvement of the GKLs' activities will require considerable assistance from the Com- mittee's institutes and republican administrations. It is advisable to organize fruitful cooperation between the workers of the Committee 's institutes and the GKLs.

It is necessary to insist that the Committee 's republican administratiom should solve simply and rapidly the problems confronting them, and should supervise the work of the state inspection laboratories without bureaucratic tendencies or procrastination. The administrations should have qualified personnel capable of rendering practical assistance on the spot.

A full understanding of the outstanding social obligations and responsibilities conferred on state inspectors of measuring equipment by the USSR government, conscientiousness, honesty and cooperation, these are the moral prin- ciples by which the GKL workers should be guided in their activity in order to be equal to the tasks set them by the Communist Party and the Soviet people,

D E F I N I N G THE C O N C E P T OF M E A S U R E M E N T *

K. B. K a r a n d e e v , V. I . R a b i n o v i c h , a n d M.

Translated from Izmeritel 'naya Teklmika, No. 12, pp. 4-6, December, 1961

P. T s a p e n k o

The theory of systems s selecting, processing, storing and supplying information in a digital form, that of measuring information systems [1], is now being established. It seems natural to use the theory of information methods for analyzing and synthesizing automatic systems of this type. However, it is hardly possible to transfer mechanically from the theory of information to that of measurements the concepts and terms which have received a specific signi- ficance in the first-named theory. On the other hand, some of the existing metrological concepts require verification in view of the development of a new trend in the theory of measurements. In the first instance this applies to the con- cept of measurement, whose definition was formulated at a t ime when the basic means of measurement consisted of pointer and manually operated instruments.

In this article we provide an anlysis of the existing definition of measurement, and suggest and justify a new in- terpretation of this concept.

I. It is generally accepted to call measurement the cognition process which consists of comparing by means of a physical experiment the measured physical quantity with a certain value of that quantity adopted as a unit of com- parison (measurement) [2, 8, etc.]. This definition has not lost its significance and, for instance, still holds for direct, nonautomatic measurements. However, in our opinion, the theory and practice of measurements definitely require clarification and partial amendment of the above definition. Let us quote a few considerations which confirm the above opinion.

1) The rapidly increasing requirements for the automation of production and experimental investigations (inthe wider meaning of the word) entail a rapid development of automatic measurements. In automatic measurements the operator is, to a certain extent, relieved from participation in the measurements. In many instances the measurement results are used for direct automatic control of certain processes. In such cases it is hardly possible to consider meas- urements as a cognition process, since the latter presupposes the presence and actions of a human operator.

2) In measuring instruments the readings have two basic forms.

* Contribution to a discussion.

938

a) a continuous form, characterized by the fact that the number of possible reading points is infinite. Such a form of indications is characteristic of pointer instruments, analog recorders, etc.;

b) a discrete form of reading, which is characterized by a finite number of points. This group includes digital measuring instruments which are based on quantization and digital coding [4]. (By quantization we understand an ap- proximation of a continuous quantity to the nearest scale division, and by digital coding the expression of the quan- tized measured quantity in a digital form.)

In measurements made by means of instruments of the analog type the quantization and digital coding are done by the human experimenter.

Thus, the measuring operations differ in instruments with different types of reading. Therefore it is difficult to answer the question when any given measurement should be considered complete. The definition of measurement which is at present adopted makes it impossible to eliminate the existing ambiguity.

It is clear that measurement is a specially prepared process whose results can be used without any additional operations. A measurement result which can be used directly consists of a numerical expression of the measured value. Such a posing of the problem in each specific case provides a single-valued answer to the question whether the process has been completed or not. Moreover, it becomes possible to define precisely the instruments in which the measuring process is completely automatic. This group includes only automatic digital measuring instruments in which all the required operations are made without the participation of a human operator.

I11 defining the term measurement it is advisable according to the above reasoning to note that the measure- ment result must be presented in a digital form.

3) A clearly distinguishable tendency in modern measurement techniques consists of raising the relative im- portance of indirect and combined measurements. In such methods the unknown variable has a functional relation- ship to the variables whose values can be obtained by direct measurements. The final result is obtained by determin- ing the functional relationship, thus involving appropriate logical and computing o~erations. In automatically oper- ating measuring instruments these operations, which are required for obtaining the result in a final form, are com- pleted without the participation of a human operator. When nonautomatic instruments are used the logical and com- puting operations must be performed by the experimenter himself. Let us note that in certain measuring instruments (for instance, electrodynamic and moving-coil wattmeters) the devices which perform the comparison and computa- tions have, for a long time, been constructed in a combined common unit.

In view of the above reasoning it should be pointed out that measurements must include the required logical and computation operations.

4) We also consider it completely incorrect to use the adjective "physical" for indicating the material nature of the experiment and the compared quantities. It seems to us that comparison may include, physical, chemical and other quantities. The same reasoning may be applied to the types of experiments performed. It is, therefore, inad- visable to indicate in the definition of measurement the nature either of the measured quantity ot the experiment.

In view of the above reasons we suggest the following definition: "Measurement is a process of obtaining in- formation which consists in comparing experimentally known and unknown quantities or signals, in performing the required logical and computing operations and presenting the information in a digital form."

Let us make additional explanations of some of the peculiarities of this definition.

The term comparison is used in the above definition in the sense of the juxtaposition of similar quantities, namely, measured and known quantities. The values of these quantities can be compared by performing the required number of subtracting or dividing operations. The comparison method or technique can vary considerably. It is also asserted that the comparison must be made by a purely experimental method which may include either the measured quantities or the signals corresponding to them and obtained by means of intermediate transformations.

Let us also note that the comparison need not be simultaneous [3]. In such a case a certain reference quantity (measure) is used either for calibrating an auxiliary quantity which is employed in direct measurements,or for replacing (substituting) the measured quantity. The substitution method of measurements consists of comparing the effects pro- duced by the compared quantities or the signals corresponding to them.

It should also be noted that the term "comparison" is not basic and its interpretation may not be single-valued. In this connection we may require in future a more detailed analysis of this concept.

939

Finally, it is suggested that the measurement be considered as completed if the measured information is pre- sented in a digital form.

It appears to us that the above formulation is suitable for characterizing direct, indirect and combined, auto- matic and nonautomatic measurements, and makes it possible to determine whether the measurement process has been completed. This definition also shows that it is impossible not to take into consideration the information character- istics of the measured systems.

II. The suggested definition of the concept of measurement includes the terms information and signal, which have such a wide content that they can be considered as philosophical terms. There is, however, no accepted defini- tion of the above terms.

A probability approach [5-10 and others] seems the most justified in dealing with the concept of information. Such an approach provides a sufficiently wide interpretation of this concept for characterizing a most varied range of phenomena and conditions, and for introducing a measure of the quantity of information. However, the authors of the above works do not consider a quantitative measure of the value of information to be essential. It is difficult to agree with this opinion, since an analysis of information received from the phenomena under investigation is of p~ramount interest. The introduction of a measure of the value of information will provide new possibilities for investigating the measurement results and designing measuring systems to be optimum in the widest meaning of the word. The publi- cation of works [11, 12] provides the hope that the required criteria will be found.

The probability nature of information makes it impossible to formulate its definition without using the terms accepted in the theory of probability and in mathematical statistics. Experiment is one of these terms. Experiment is defined as the implementation of certain conditions and actions. Such an interpretation provides the possibility of examining any happenings of practical interest. Event is another necessary term. In the theory of probability event is the qualitative result of a n~xperiment. In defining the concept of information we must, in addition to the above considerations, also reflect the following two circumstances.

1) "Information" is undoubtedly a collective term for denoting the contents of the studied processes and pheno- mena independently of their nature. Moreover, it is assumed that the content of the phenomena and processes may be of a most diverse nature. If there is no diversity, no multiplicity of possibilities, the use of the term information cannot be justified. Hence, the possibility of selection is prerequisite for obtaining information. The information contains the qualitative results of this selection, i.e. implemented events. Hence, in defining the term "information" we must deal with the content of events.

2) It is essentially important to know what constitutes the difference in an experiment before and after the con- tents of an implemented event become known. The answer is obvious, the initial indeterminacy of the experiment changes. This reflects the essence of information. In our opinion it would be incorrect to consider only a reduction in the indeterminacy. Such an approach would impede the establishment of a measure of the value of information, since a content of events is possible which will lead to a rise of indeterminacy in an experiment.

On the basis of the above considerations it is possible to make the following definition: "Information denotes a content of events which have an initial experimental indeterminacy."

This definition is in complete agreementwith the formulas adopted for calculating the quantity of information. At the same time it does not contradict the fact that the value of the same amount of information may differ.

Let us now describe the considerations which were adopted in formulating the definition of the concept signal.

There is no doubt that a necessary condition for selecting information from a process under consideration and the storing and transmission of this information is provided by a property of matter which approaches perception, namely, a property of elementary mapping. The phenomena which occur in the mapping process or arise as a result of it contain information on the causes which have produced this process. These phenomena provide the information, since they make it perceptible. These phenomena are denoted by their information content as sisnals.

Let us now note the most important characteristics of the phenomena which occur in mapping processes, namely, of signals. Under appropriate conditions these phenomena can remain invariable for an indefinite time. It is precise- ly these properties of signals which are used in "memory" devices. Another equally important property of signals con- sists of their ability to react on certain objects of the material world. These reactions produce new phenomena which

940

are also signals, etc. Without such a chain of reactions it is impossible to consider the propagation of signals, and, hence, the transmission of information.

The above considerations have made it possible to formulate the following definition: "A signal is a pheno- menon which serves to transmit information by means of reactions."

In conclusion let us note that measuring devices are a variety or a part of information systems. The extension of the sphere of application and the raising of performance requirements of automatic measuring information systems require a search for new methods of their analysis and synthesis. The application of such information theory character- istics as entropy, carrying capacity, redundancy, etc. is very convenient in studying the potentialities of measuring information systems by comparing them to each other and determining their approximation to an optimum design. In order to develop rapidly the theory of measuring information systems it is necessary to use and appropriately evolve the techniques of the information theory, the theory of algorithms, and linear and dynamic programming.

1B 2. 3. 4. 5.

6.

7. 8. 9. 10. 11. 12.

L ITERATURE C I T E D

K. B. Karandeev, Vestnik AN SSR, 1961, No. 10. M. F. Malikov, Foundations of Metrology, part I, 1949. K. B. Karandeev, Elektrichestvo, 1949, No. 7. M. P. Tsapenko, Izmeritel 'naya tekhnika, 1961, No. 5. K. Shannon, Mathematical Theory of Communications, in collection "Theory of Transmitting Messages in the Presence of Interference " [Russian translation] (Foreign Literature Press, 1953). K. Shannon, Communication in the Presence of Noise, in eolleetinn "Information Theory and Its Applications" [Russian translation] (Fizruatgiz, Moscow, 1959). S. Goldman. Theory of Information [Russian translation] (Foreign Literature Press, 1957). A. A. Fel'dbaum, Calculating Devices in Automatic Systems [in Russian] (Fizmatgiz, Moscow, 1959). A. A. Kharkevieh, Outline of a General Communications Theory [in Russian] (GITTL, Moscow, 1955). A. M. Yaglom and I. M. Yagloru, Probability and Information [in Russian] (Fizmatgiz, Moscow, 1960). A. A. Kharkevich, Problemy kibernetiki, 1960, No, 4. G. J. Schouter, Collection entitled "Theory of Message Transmission" [Russian translation] (Foreign Literature Press, 1957).

D E T E R M I N I N G IN S T A T E T E S T I N G WHETHER I N S T R U M E N T S ARE

RELIABLE AND S U I T A B L E FOR MASS P R O D U C T I O N

M. A. Z e m e l ' m a n and N. I. T y u r i n

Translated from Izmeritel"naya Tekhnika, No. 12, pp. 7-8, December, 1961

In state testing of experimental models of measuring instruments a number of important questions need to be solved, for instance, whether the instrument meets the requirements of our national economy; whether its principle of operation, circuit and construction are up-to-date; and whether it ensures uniformity of measurements under vari- ous specified conditions of operations. The reliability of instruments and the possibility of their mass production are also important questions.

Under capacity for mass production we understand the properties of the instruments' circuits and construction which ensure the required quality of instruments when they are assembled from components which are made according to definite technical specifications and drawings and meet the tolerances specified for them. No theory has as yet

941